Visual perception depends on more than visual input. The brain constantly generates internal signals that modulate responses to external visual stimuli. Dynamic internal signals allow neural networks to respond differently to the same stimulus, enabling organisms to adapt to changing task demands. Internal network dynamics also support persistent neural activity related to visual working memory. In this proposal, we examine the influence of internal signals, continuing our focus on parietal cortex. An important class of internal signals are related to self-movement. We have identified a powerful suppression of the firing of parietal neurons by microsaccades - but only during a task requiring a transient motion-detection that may be mimicked by a microsaccade's visual effect. Aim 1 examines the hypothesis that the suppression may be flexibly gated by task demands. Aim 1 also investigates the potential role of the rostral superior colliculus, an area involved in microsaccade generation, in the suppression. Aim 2 tests whether two different neuronal signals that have been reported in parietal cortex -- categorization and perceptual decision -- are actually the same thing. Most perceptual decision experiments did not test whether the signals could be independent of the form of the animal's report; we propose that if perceptual decision signals were independent of the report, they would be indistinguishable from categorical signals. We will examine whether LIP neurons provide generic visual categorical signals, and we will compare perceptual decision and categorical signals head-to- head, in the same experiment, in the same neurons. Aim 3 examines the mechanisms underlying persistent firing in parietal cortex. Recurrent network models posit that persistent memory-delay period activity is a natural dynamic of recurrent circuits. A recent recurrent network model makes several robust yet surprising predictions; for example, if different levels of persistent activity can be evoked under different conditions, the amplitude of the activities across the neuronal population should be scaled versions of each other and of the spontaneous activity. In addition, there should be a common order of preference among neurons in their persistent firing, a highly unusual organization for cortex. Aim 3 uses the experimental data generated from Aims 1&2 to test and refine this model. Our experiments will provide a mechanistic perspective on how internal states of the brain facilitate and affect the perception of the external visual world. Gaining basic information on these processes is an essential step for understanding normal and abnormal brain processing.